We have all heard people in government and the nonprofit sector talk up the idea of data driven decision making, but the truth is this is more often than not just talk and not backed up by real action. So when local government really takes this seriously and does actively use data for planning, decision making and the like we’re really excited about it. In Oakland we helped the city develop this thing called a Stressor index many years ago for just these purposes. People wanted to know some relative measure of stress or distress on a neighborhood compared to others in the city, and as this was largely related to public safety we chose police beats as the areas for this model. We took readily available census data on populations, poverty, crime, arrests, public assistance, school issues and more and built a simple but useful model of the stressors a neighborhood can face. it’s been very useful for both the city, the Oakland Police Dept (OPD) and many nonprofits involved in violence prevention and community development.
With the arrival of the new census 2010 data for our cities and neighborhoods this year the City asked us to update this useful model so as to be making decisions based on the latest numbers- an important move given the massive changes in Oaktown as a result of violence, foreclosures, massive unemployment, school problems and the rest- our city has changed and so must our data right! Cue disaster. Of the federal kind. Our all-knowing government has totally revamped our census process, and the first of the measurable impacts of their decisions are just hitting home in Oakland. The Stressor model, considered very important to many agencies is dead. With the limited information collected in the 2010 Census, the bureau is unable to produce most of what was published using the 2000 census, including the all important measures of poverty, population on public assistance and unemployment. Pretty important measures for this model. There is always the American Community Survey (ACS) to consider as a data source- unfortunately because of poor/small sampling the data are only being released at the census tract level and no lower- and the tract boundaries don’t even come close to matching the beat boundaries (or should that be the other way around…?). Given a huge spatial misalignment and the massive margins of error in these ACS data (as these are sampled data they have errors- at the tract level the margins of error for poverty in Oakland are as large as the numbers themselves- very hard to trust!), we no longer have any way to calculate these essential data for the Stressors model.
As opendata and opengov are increasingly powerful movements, without quality data of the right type even this new Gov2.0 effort often is not enough. When federal agencies do not have a grasp on the needs and uses of their data at a grassroots, academic and community level they will make decisions that impact us and as with the ACS these decisions are now starting to have a serious impact on important work in our communities. Data driven is great, but poor data means poor decisions.
We’re open to ideas others have about rebuilding the Stressor model using other sources, we have some ideas but would love your thoughts too!